The Agent Skills Directory

[SAFE]: The skill contains legitimate instructions for assisting users with Malloy model development, including proposing source architectures and defining dimensions and measures based on data evidence.
[INDIRECT_PROMPT_INJECTION]: The skill defines a workflow for ingesting model metadata and querying data via get_context and execute_query. While these tools represent an entry point for external data, the skill guides the agent to perform specific analytical tasks (like calculating distributions and identifying grain) which are non-executable in nature and typical for this domain.
[DATA_EXPOSURE]: The skill uses database query tools to inspect schema and distributions. This behavior is confined to the expected scope of a data modeling agent and does not include any patterns of data exfiltration or hardcoded credentials.

malloy-define